Enzyme therapy and immune response in relation to CRIM status: the Dutch experience in classic infantile Pompe disease

Patients

We describe 11 patients with classic infantile Pompe disease who received ERT in our hospital between 1999 and 2012. Classic infantile Pompe disease was defined as symptoms of muscle weakness within 6 months after birth; hypertrophic cardiomyopathy (left ventricular mass index (LVMI) > +2SD (>75 g/m²) (Poutanen and Jokinen 2007)); less than 1 % acid α-glucosidase activity in fibroblasts; and severe mutations in both GAA alleles. Activity assays and mutation analysis were performed as described previously (Kroos et al 1997, 2007). All patients participated in consecutive trials investigating the safety and efficacy of ERT (20 mg/kg every other week to 40 mg/kg weekly). Initially, four patients received recombinant human α-glucosidase from the milk of transgenic rabbits (Van den Hout et al 2000). From 2004 onward, all patients were treated with alglucosidase alfa. The Institutional Review Board approved all studies and the parents of all patients gave written informed consent.

CRIM status

Two methods were used to determine the patients’ CRIM status. Cultured skin fibroblasts from the patients were used in the first method. They were grown in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10 % fetal calf serum (FCS) and antibiotics, and were harvested with trypsin 1 week after reaching confluence. Frozen cell pellets were lysed in 10 mM phosphate-buffered saline (PBS) pH 7.0 containing 1 % Triton-X100 and protease inhibitors. A 10 μL aliquot containing 100 μg protein was mixed with 10 μL sample buffer (NuPAGE, LDS sample buffer, Life Technology, with 4 % SDS instead of 0.84 %) and applied to SDS-PAGE (Bio-Rad Criterion XT 4–12 % MOPS gel system). After blotting onto nitrocellulose the various molecular forms of acid α-glucosidase were visualized with polyclonal rabbit antibodies against recombinant-human acid α-glucosidase and goat anti-rabbit IRDye 680LT as secondary antibody using an Odyssey infrared imager (LI-COR Biosciences).

The second method for determining the CRIM status was based on transient expression of GAA cDNA constructs containing the patients’ pathogenic mutations in HEK293T cells (one mutation per construct). The activity of acid α-glucosidase in these cells and in the culture medium was measured with the artificial substrate 4-methylumbelliferyl-α-D-glucopyranoside (MUGlc, Sigma) at 72 h after transfection. The biosynthetic forms of acid α-glucosidase were separated by SDS-PAGE and visualized by immunoblotting (Hermans et al 1998).

Antibody detection

Blood samples for antibody titer measurements were drawn before the start of ERT and every 3 months thereafter just before the start of enzyme infusion. The sera were stored at -20°C. An enzyme-linked immunosorbent assay (ELISA) was used to determine the titers. To this end a 96-well plate (Nunc, F96 Maxisorp, Denmark) was coated with 50 μL/well alglucosidase alfa in a concentration of 5 μg/mL, diluted in PBS (pH 7.4), and incubated for 2 h at room temperature under continuous shaking. Plates were blocked overnight at 4°C with a solution of bovine serum albumin (BSA; Sigma A7030) in PBS (1 % weight/volume). The plates were rinsed six times at room temperature with 200 μL washing buffer (0.05 % Tween-20 in PBS). Plates were then incubated with 50 μL of five-fold serial dilutions of patients’ sera for 1 h during shaking. Samples were diluted from 50 to 3,906,250 fold in dilution buffer (BSA/PBS containing 0.05 % Tween-20). Sera from healthy individuals were used as negative controls, and rabbit antiserum against alglucosidase alfa as a positive control. After washing of the plates, 50 μL conjugate was added to each well: polyclonal goat-anti-human-[IgG, IgA and IgM]-HRP (Acris) in a 20,000-fold dilution for human sera, and goat-anti-rabbit-IgG-HRP (Sigma) in a 10,000-fold dilution for rabbit serum. After 1 h incubation at room temperature followed by thorough washing, 100 μL Tetramethylbenzidine Microwell Peroxidase substrate (Kirkegaard and Perry Laboratories, Maryland) was added, and the plates were incubated for 10 min. The colorimetric reaction was stopped by the addition of 100 μL 1 M phosphoric acid (H₃PO₄). Absorbance was measured at 450 nm with a spectrophotometer (Thermo Electron Corporation, Vantaa, Finland). The maximal dilution at which absorbance was at least twice that of the negative control was taken as titer.

An antibody titer assay based on immunoprecipitation of alglucosidase alfa was used as second semi-quantitative method (de Vries et al 2010).

Clinical assessments

The clinical outcome was evaluated by analyzing survival, ventilator use, gross motor function, and cardiac dimensions at baseline and at regular intervals; the latter by 2D-guided M-mode echocardiography. The LVMI was calculated as a measure for hypertrophic cardiomyopathy. Motor development was assessed using the Alberta Infant Motor Scale (AIMS) (Piper and Darrah 1994).

Safety assessments included the monitoring of infusion-associated reactions (IARs). All adverse events that were judged to be possibly, probably or definitely related to ERT were considered IARs.

Statistical analysis

Due to limitations in patient numbers, the clinical data was summarized using descriptive statistics. All patients were followed-up for at least 4 years after the start of ERT, or until death. Data were analyzed using SPSS for Windows version 20, SPSS Inc., Chicago, IL.

Patients

This study describes 11 patients (five boys and six girls) with classic infantile Pompe disease (Table 1). All had symptoms before the age of 3.5 months; nine started ERT before the age of 4 months. Two patients started ERT at the ages of 7 and 8 months when they were severely affected. All 11 patients had cardiac hypertrophy, less than 1 % of average normal acid α-glucosidase activity in cultured fibroblasts, and a severe mutation in both GAA alleles (Table 1). They were treated with ERT for 0.3 to 13.7 years (median duration 5.6 years).

CRIM status

‘CRIM status’ in this study is defined as the ability to detect any acid α-glucosidase with immunological methods in either the patients’ cultured fibroblasts or in HEK293 cells in which mutated GAA constructs with the patients’ mutations were expressed. A CRIM-positive status indicates that molecular forms of acid α-glucosidase could be detected. A CRIM-negative status indicates that these were undetectable.

Figure 1 presents an example of the first type of analysis in fibroblasts. Cell homogenates were analyzed by SDS-PAGE followed by immunoblotting. Compared to normal synthesis of acid α-glucosidase in fibroblasts of a healthy individual (Wt), represented by the presence of a 110 kD precursor and molecular species of 95, 76 and 70 kD, fully maturated enzyme of 76 kD was not detected in cells from any of the patients. However, the 110 kD precursor was detected in eight cases. In four of these cases (pts 1, 3, 10, and 11), the apparent molecular mass of the precursor was near normal. In the other four cases (pts 2, 5, 6, and 9), the mass was abnormally low due to either homozygosity or heterozygosity for the rather common exon 18 deletion (c.2481 + 102_2646 + 31del). Proteolytic conversion of the 110 kD precursor to the 95 kD intermediate could be demonstrated only in cells of patient 9. Three patients did not have any trace of acid α-glucosidase due to either homozygosity for c.525delT (pt 7) or c.2741delinsCAG (pt 4), or to compound heterozygosity for c.525delT and c.378_379del (pt 8). Based on the outcome of this analysis, eight of the 11 patients were designated CRIM-positive and three CRIM-negative.

As an alternative approach determining the patients’ CRIM status, we introduced the mutations in a GAA cDNA expression construct and studied their isolated effect on acid α-glucosidase synthesis and maturation in HEK293T cells (Fig. 2). Transient expression of the five missense mutations identified among the patients in this study revealed synthesis of the 110 kD acid α-glucosidase precursor with either normal (c.1933G > T, c.1913G > T, and c.1115A > T) or slightly higher than normal (c.1460 T > C and c.1799G > A) molecular mass. The c.1799G > A and c.1460 T > C encoded precursors were also clearly present in the culture medium, but the c.1933G > T, c.1913G > T, and c.1115A > T encoded precursors were not (Fig. 2; medium). Figures 1 and 2 both illustrate that the conversion from 110 kD to 95 kD is not blocked completely in the c.1799G > A construct; even the 76 kD form was detectable.

Transient expression of c.2741delinsCAG did not reveal any acid α-glucosidase species (Fig. 2; cells) and also this finding is consistent with the lack of acid α-glucosidase expression in fibroblasts of the patient who is homozygous for this mutation (Fig. 1). None of the mutagenized cDNA constructs led to expression of acid α-glucosidase activity in the transfected cells nor in the medium.

On the basis of these two types of analyses, patients 4, 7, and 8 were designated CRIM-negative; the other eight patients were designated CRIM-positive.

Antibody titers

All patients developed antibodies to alglucosidase alfa (Supplemental Fig. 1a, Table 2). Over 4 years of treatment the median peak antibody titer (ELISA) was 1:31,250 (range 1:1250–1:156,250). These titers were reached within 1 year of ERT. Both low and high antibody titers were measured in CRIM-positive (median 1:18,750, range 1:1250–1:156,250) and CRIM-negative patients (median 1:31,250, range 1:6250–1:156,250). The peak antibody titers tended to be higher in patients who had started ERT relatively late (Table 2). Notably, none of the patients who received ERT before the age of 2 months developed titers above 1:6250.

Six patients (pts 5–8, 10 and 11) developed titers ≥ 1:31,250 (Table 2); titers remained high in five cases and declined temporarily to 1:6250 in one case (pt 10). The antibody titers of the other five patients remained ≤ 1:6250 over the entire study period; one of these patients was CRIM-negative and his titer eventually declined to 1:50. There was no apparent relationship between the titer and the dose of ERT, nor between the titer and the use of recombinant human acid α-glucosidase from either rabbit milk or CHO cells (results not shown).

An immunoprecipitation assay was used as an alternative method for determining the antibody titer, for which we selected three patients whose respective ELISA titers were 1:1250, 1:6250, and 1:156,250. The results are shown in Supplemental Fig. 2 and are consistent with the outcome of the ELISA assay in that both methods discriminate similarly between a low, intermediate and high titer.

Clinical outcome over 4 years of ERT

GAA genotypes and CRIM status

The quality and quantity of acid α-glucosidase in Pompe disease depends on the GAA genotype. Ten different mutations were identified, all causing complete loss of acid α-glucosidase activity.

The commonly used procedure to determine the CRIM status is immunoblotting and normally reveals four distinct acid α-glucosidase species that separately and collectively are called CRIM. In healthy people (Fig. 1, Wt), the long-lived 76 and 70 kD species comprise approximately 80 % of the total CRIM. In patients, however, the synthesis of acid α-glucosidase can be totally lost, which results in a CRIM-negative status. More often, the synthesis is derailed, and CRIM consists mainly of the 110 kD precursor (e.g., Fig. 1, pts. 1 and 3) or a modified derivative thereof (Fig. 1, pts. 2 and 6). As it stands, the CRIM-negative status is well defined, but the CRIM-positive status encompasses a collection of qualitatively and quantitatively abnormal immunoblot profiles.

By and large, a patient’s CRIM status can be predicted on the basis of its genotype (Bali et al 2012). Most non-sense mutations and frame-shift mutations lead to mRNA decay and a CRIM-negative status (e.g., 525delT in pt. 3), truncated protein species are sometimes detectable by immunoblotting (e.g., c.2481 + 102_2646 + 31del in pts. 1, 7, 10, and 11), and thus lead to a CRIM-positive status. Although the effect of missense mutations is hard to predict, they usually lead to a CRIM-positive status.

Information on the effect of missense mutations can be obtained by transient expression (Fig. 2), which is especially informative for the stage at which mutant forms of acid α-glucosidase are degraded. For instance, the c.1460 T > C or c.1799G > A encoded acid a-glucosidase precursors must have traversed the Golgi complex since they are secreted into the medium, whereas the c.1833G > T, c1913G > T and c1115A > T encoded precursors do not appear in the medium (Fig. 2) and are apparently degraded while passing through the ER/Golgi complex. By the current definition of CRIM, these five missense mutations lead to a CRIM-positive status, but obviously do not represent one and the same CRIM-positive condition.

Our application of both methods to determine the CRIM status resulted in the same outcome: eight patients are CRIM-positive and three are CRIM-negative. Notably, the patient homozygous for c.2741delinsCAG was previously designated CRIM-positive (Hermans et al 1998; Banugaria et al 2011).

CRIM status and antibody titer

Antibody formation is a natural response to foreign invading proteins. Thus, ERT is prone to evoke an immune response in CRIM-negative patients, but not necessarily in all CRIM-positive patients. Patients with Pompe disease receiving ERT respond roughly according to these principles. Analysis of the immune response in 34 treated infants showed a strong tendency toward higher and sustained antibody titers in CRIM-negative compared to CRIM-positive infants (Kishnani et al 2010). However, CRIM-positive patients can also develop high titers (Banugaria et al 2011), and CRIM-negative patients low titers (Abbott et al 2011; Al Khallaf et al 2013). Our study in 11 infants has led to similar findings. The highest peak antibody titers were measured in four of the eight CRIM-positive patients and in two of the three CRIM-negative patients. The antibody titer of the third CRIM-negative patient did not exceed 1:6250 and spontaneously regressed to 1:50 after 3 years of ERT. There were two CRIM-positive patients with the same GAA genotype; one developed a relatively high titer, and the other a relatively low titer.

Although the number of patients in our study is small, we can firmly conclude that the CRIM status alone predicts neither the level nor the duration of the immune response. This may relate to the imprecise definition of CRIM status, which does not describe the amount, the conformation or the location of the endogenous acid α-glucosidase. For instance, secretion of the 110 kD precursor as opposed to intracellular degradation might reduce the immune response and contribute to the relatively low antibody titers in patients 1 and 9. Notably, adult patients with a considerable amount of normally structured and catalytically active acid α-glucosidase can also develop high titers (van der Ploeg et al 2010; de Vries et al 2010; Patel et al 2012; Regnery et al 2012). Altogether, we conclude that genotype and CRIM status are of limited value in predicting the height of the immune response.

As previously suggested (van Gelder et al 2012; Al Khallaf et al 2013), our study indicates that the age at start of ERT might play a role in the immune response since none of the patients who started ERT before 2 months of age developed titers > 1:6250. There are at least two plausible explanations for this: First, the neonatal immune system is immature, and very early administration of ERT might induce tolerance (Brooks 1999; Dierenfeld et al 2010). Second, the ‘danger model’ suggests that the immune system needs alarm signals from injured tissues to be activated. At a less advanced stage of disease these signals will be weaker (Matzinger 2002). Other models may equally apply.

Consequences of antibody formation

Depending on their binding sites antibodies can inhibit catalytic function, block binding to the mannose 6-phosphate receptor and prevent uptake, or otherwise misdirect the protein to macrophages and neutrophils (Brooks et al 2003; Wang et al 2008). Antibodies and antigen can also form immune complexes and trigger a cascade of adverse events (Hunley et al 2004).

In a previous case-report about an adult patient with Pompe disease we have calculated how the concentration of alglucosidase alfa specific antibodies relates to the concentration of alglucosidase alfa during enzyme infusion (de Vries et al 2010). In the present study we have applied the same arithmetic method to infants. For example, patient 11 had an ELISA titer of 1:156,250 at a certain point of treatment. At that time the corresponding titer as measured by immune-precipitation was 40 nmol MUGlc/h.μL (Supplemental Fig. 2) implying that 1 mL of the patient’s serum contained enough antibodies to bind 0.13 mg alglucosidase alfa (de Vries et al 2010). As a child receives 0.24 mg of alglucosidase alfa per mL blood (based on 20 mg/kg and a circulating blood volume of 80 mL/kg), antibodies may bind as much as 54 % of the administered enzyme. By contrast, if the titer is only 4 nmol MUGlc/h.μL, corresponding with an ELISA titer of 1:6250, the circulating antibodies can bind only 5 % of the administered aglucosidase alfa.

Although these estimates are crude they indicate that ELISA titers of 1:6250 and lower probably have no clinical significance, whereas titers of 1:31,250 and higher may counteract ERT at a dose of 20 mg/kg. Accordingly, titers above 1:60,000 are expected to counteract ERT at a dose of 40 mg/kg. Our arithmetic estimates of what should be considered a high titer and what should be considered a low titer remarkably correspond to the cut-off value of 1:51,200 chosen by Banaguria et al (Banugaria et al 2011).

Antibodies can impede the effect of ERT in several ways. In four cases we observed a decrease of catalytic activity suggestive for binding of antibodies to the enzyme’s active site. Uptake was inhibited in three cases. In one case this was due to a combination of inactivation and uptake inhibition possibly by steric hindrance of the ligand-receptor binding. Of note, the effect of antibodies not only depends on the binding sites but also on the stoichiometry of antibodies and alglucosidase alfa in the experimental setting; changing the concentration of one of the two components can tip the balance. Whereas neutralizing antibodies were reported to be more common in CRIM-negative than in CRIM-positive patients (Kishnani et al 2010; Banugaria et al 2011), in our study their presence was not related specifically to the patients’ CRIM status but to the patients’ antibody titer.

It has been shown that high antibody titers in CRIM-negative as well as CRIM-positive patients are associated with shorter ventilator-free survival (Kishnani et al 2010; Banugaria et al 2011). In several adults with Pompe disease, a high titer was also accompanied by poor response to ERT (de Vries et al 2010; Patel et al 2012). Although our study group was too small and too varied to permit statistical analysis, patients with higher titers tended to attain fewer motor milestones. However, with respect to ventilator-free survival, only two of the six patients who developed respiratory insufficiency or died had a titer of ≥ 1:31,250 at the time that these events occurred. Two other patients had low titers, but nevertheless developed respiratory insufficiency at the age of 2. Two patients became ventilator dependent just before or soon after the start of ERT which we ascribe to their advanced disease at the start of ERT.

Conclusions and challenges

The outcome of classic infantile Pompe patients is determined by many factors, among which are the age and disease severity at initiation of ERT, the patient’s genotype, and the height of the antibody response. It is beyond question that high and sustained antibody titers need to be prevented in order to achieve a good response to ERT (Kishnani et al 2010; de Vries et al 2010; Banugaria et al 2011). Our study indicates that the negative effect of antibodies starts at an antibody titer of approximately 1:31,250 when the recommended dose of 20 mg/kg is used. Very early start of ERT (<2 months), achievable by neonatal screening, may help to keep the antibody titers low. Recent publications indicate that immune tolerance can be successfully induced by a combination of rituximab, methotrexate, and intravenous immunoglobulins (Mendelsohn et al 2009; Messinger et al 2012; Banugaria et al 2013b), a regimen likely to be most effective when used prophylactically (Lacaná et al 2012; Banugaria et al 2013a). As the patients’ genotype and related CRIM status alone are obviously not the sole determinants for the patients’ immune response, identification of other triggering factors remains a challenge. Though the number of patients is very small, it occurred to us that the three CRIM-negative patients in our study had an early demise while all CRIM-positive patients survived. If the difference is not made solely by the CRIM-status or antibody titer, might it be that CRIM-positive patients can have a tiny bit of acid α-glucosidase activity in their tissues that we cannot detect with our current methods?

Below is the link to the electronic supplementary material.